## 1. Precise restatement and assumptions The claim is an **existence claim about a particular reduced-form model**, not a general proposition about redistribution. Let \(d_{A,t},d_{B,t}\) be the households’ disposable incomes after any transfers in period \(t\). The next period’s pre-transfer earnings are \[ e_{A,t+1}=f(d_{B,t}),\qquad e_{B,t+1}=1, \] where \[ f(y)=5e^{-k(y-1)},\qquad k=\frac{\ln5}{2}. \] The policies being compared are: - **No redistribution:** \(d_{i,t+1}=e_{i,t+1}\). - **Complete equalization:** earnings are pooled and divided equally: \[ d_{A,t+1}=d_{B,t+1} =\frac{e_{A,t+1}+e_{B,t+1}}2. \] Relevant assumptions and definitions are: - Time is discrete, and the externality operates with a one-period lag. - A’s earnings depend on B’s **disposable income**, not B’s own earnings. - Redistribution has no administrative or direct resource cost, but can affect future production through that externality. - There are no additional behavioral responses, saving decisions, or equilibrium conditions. - “Inequality is zero” refers to **post-transfer income inequality**. - “Average total income” means the long-run arithmetic time average of aggregate earnings. Budget-balanced transfers leave each period’s aggregate disposable income equal to aggregate earnings. - “Attracting cycle” need not mean that every initial condition lies on, or converges to, that cycle. Precisely stated, the claim is that this model has a globally finite-time stable no-transfer steady state with total income \(6\), while equalization creates an attracting period-two orbit with totals \(2\) and \(6\), and hence mean total income \(4\). ## 2. Checking and independently deriving the result ### Without redistribution The update map is \[ N(a,b)=(f(b),1). \] Applying it twice gives \[ N^2(a,b)=(f(1),1)=(5,1). \] Thus every admissible initial income vector reaches \((5,1)\) within two updates. Any fixed point must have B’s income equal to \(1\), and therefore A’s equal to \(5\), so the fixed point is unique. Steady-state aggregate income is \[ 5+1=6. \] This part is correct. ### With complete equalization After pooling, write both disposable incomes as \(y_t\). Then \[ y_{t+1}=g(y_t) =\frac{1+5e^{-k(y_t-1)}}2. \] Because \[ e^{-2k}=e^{-\ln5}=\frac15, \] we obtain \[ g(1)=3,\qquad g(3)=1. \] Thus \(1\leftrightarrow3\) is an exact period-two orbit. In household disposable incomes, it is \[ (1,1)\leftrightarrow(3,3). \] The corresponding pre-transfer earnings vectors alternate between \[ (5,1)\quad\text{and}\quad(1,1), \] so aggregate earnings alternate between \(6\) and \(2\). A particularly natural initialization makes the comparison exact: start at the no-policy steady state \((5,1)\), then pool its income. This produces disposable incomes \((3,3)\), followed by \((1,1)\), and so on. ### Attraction Differentiate: \[ g'(y)=-\frac{k}{2}f(y). \] Hence \[ g'(1)=-\frac{5k}{2},\qquad g'(3)=-\frac{k}{2}. \] The derivative of the two-step map at either cycle point is \[ (g^2)'(1)=(g^2)'(3) =g'(1)g'(3) =\frac{5k^2}{4} =\frac{5(\ln5)^2}{16} \approx0.80947. \] Its absolute value is strictly below one, proving local attraction. Notice that one individual step has slope magnitude above one; stability depends on the **product over the full cycle**, which the text correctly uses. ### Average income On the cycle, \[ \overline{Y}=\frac{2+6}{2}=4. \] Relative to the no-transfer benchmark, \[ \frac{6-4}{6}=\frac13. \] All stated calculations check out. ## 3. Trying to break the result ### Important exception: equalization also has a steady state Since \(g\) is continuous and strictly decreasing, \(g(y)-y\) is strictly decreasing. Because \[ g(1)-1=2>0,\qquad g(3)-3=-2<0, \] there is exactly one fixed point \(q\in(1,3)\), numerically \[ q\approx1.80646. \] If initialized exactly at \(q\), the equalized economy remains there forever: there is **no cycle**. Its aggregate income is about \(3.613\), not \(4\). Thus a universal assertion that *every* equalized trajectory cycles and averages \(4\) would be false. The actual existential claim survives this exception. ### A stronger, complete basin result In fact, every other admissible initial condition converges to the stated cycle. This can be proved without relying on simulation. Make the affine change of variable \[ x=k\left(y-\frac12\right). \] The equalization map becomes \[ x_{t+1}=T(x_t)=ae^{-x_t}, \qquad a=\frac{5k}{2}e^{k/2}\approx3.00834. \] The two cycle points are \[ \ell=\frac{k}{2},\qquad r=\frac{5k}{2}, \] with \(T(\ell)=r\) and \(T(r)=\ell\). Let \(p\in(\ell,r)\) be the unique fixed point of \(T\). For the two-step map \(H=T\circ T\), set \(z=T(x)\). Then \[ H'(x)=az e^{-z}. \] As \(x\) increases, \(z\) decreases continuously, and this derivative increases to one maximum and then decreases. Consequently, \(H'(x)=1\) has at most two solutions, so Rolle’s theorem implies that \(H(x)-x\) has at most three distinct zeros. We already know three: \[ \ell,\quad p,\quad r. \] They are therefore all its zeros. Moreover, \[ H'(\ell)=H'(r)=\ell r\approx0.80947<1. \] Together with continuity and the absence of other zeros, this gives the signs \[ H(x)-x: \quad +,\;-,\;+, \;- \] on the four intervals separated by \(\ell,p,r\). Since \(H\) is increasing, its iterates therefore converge monotonically: - to \(\ell\) if \(x_0
p\); - remain at \(p\) if \(x_0=p\). The intervening applications of \(T\) exchange \(\ell\) and \(r\). Thus, except for the unique equalized fixed point, trajectories converge to the period-two orbit. Their long-run mean aggregate income is consequently \(4\), even if they do not alternate exactly from the outset. ### What is—and is not—robust - **Positivity is satisfied.** Earnings are positive, and equalization yields \(y_{t+1}>1/2\). - **Transfers balance each period.** There is no missing administrative expense or accounting loss. - **The output loss is a production response**, caused by the dependence of future earnings on B’s disposable income. - **Exact alternation from a finite date is special.** Generic initial conditions approach the cycle asymptotically. - **The precise one-third loss is calibrated.** Small smooth perturbations preserve the attracting cycle locally, but generally change its endpoints and average. - **Equalization alone is not sufficient.** For example, if \(f(y)\equiv5\), equalization immediately gives \((3,3)\), with aggregate income \(6\) forever. There is no gap in the mathematical existence proof. ## 4. Novelty The underlying dynamical mechanism is classical. The transformation above reduces the model to the decreasing exponential iteration \[ x_{t+1}=ae^{-x_t}. \] Its fixed point satisfies \(p=ae^{-p}\), and its derivative there is \(-p\). The stability threshold occurs at \[ p=1,\qquad a=e, \] the familiar flip, or period-doubling, threshold. Here \(a\approx3.00834>e\). It is also equivalent, after writing \(u=x/a\), to \[ u_{t+1}=b^{u_t},\qquad b=e^{-a}, \] a classical iterated-exponential or power-tower iteration. Economically, the mechanism resembles nonlinear cobweb dynamics: a lagged negative feedback produces alternating outcomes. I cannot establish priority for this exact two-household redistribution framing or numerical calibration. But the mathematical phenomenon is a **special case of known one-dimensional dynamics**, not a new dynamical discovery. ## 5. Does the claim fairly summarize the text? **Yes, as an existence claim.** The word “can” matters: the model genuinely supplies an attracting cycle and the stated income reduction. The main interpretive cautions are: - “Costless” means no direct transfer cost, not no production consequences. - The externality is explicitly assumed rather than derived from household optimization or a market equilibrium. - Zero inequality concerns disposable income, not pre-transfer earnings. - “Boom” is relative to the trough: the high-output phase merely matches the no-transfer benchmark. - The claim provides no evidence that equalization generally has this effect in actual economies. These qualifications limit the economic inference, but do not undermine the stated mathematical construction. {"verdict":"correct","novelty":"special case or variant of known results","fair_summary":"yes","confidence":5,"one_line":"The construction rigorously establishes the claimed attracting cycle and one-third long-run income loss, but it is a calibrated reduced-form example of classical decreasing-exponential dynamics, not a general result about redistribution."}